Ethanol consumption data are presented as mean ± S.E.M. Microarray data were analyzed using the R statistical environment [35] and Microsoft Excel (2011). Only genes with a detection p value ≤0.05 and present on >80% of arrays were utilized in the analysis of each tissue dataset. Variance stabilization transformation [36] and quantile normalization [37] were used to pre-process the data in Lumi [38]. Expression value outliers were removed using Grubbs’ test with a critical value of 2.21 or 2.29, depending on the number of analyzed arrays. One sample (a Chronic PFC control) clustered separately in the Lumi outlier detection tree and had more than 5% outlier genes and was thus removed from the analysis. Limma [39] was used to fit a linear model for each gene and detect differentially expressed genes using an empirical Bayes method. Fold changes in gene expression are given as change in treated relative to control. Significant overlap of differentially expressed genes among pairs of studies was assessed with a Bonferroni-corrected Chi-square goodness of fit test. The Pearson product-moment correlation was used to evaluate correlation of ethanol consumption with individual gene expression values.
